A comparison of predicted and observed ocean tidal loading in Alaska

We investigate the elastic and anelastic response of the crust and upper mantle across Alaska to mass loading by ocean tides. GPS-inferred surface displacements recorded by the Plate Boundary Observatory network are compared with predictions of deformation associated with the redistribution of ocean water due to the tides. We process more than 5 yr of GPS data from 131 stations using a kinematic precise point positioning algorithm and estimate tidal contributions using harmonic analysis. We also forward calculate load-induced surface displacements by convolving ocean-tide models with load Green's functions derived from spherically symmetric Earth models. We make the comparisons for dominant tidal harmonics in three frequency bands: semidiurnal (M2), diurnal (O1) and fortnightly (Mf). Vector differences between predicted and observed ocean tidal loading (OTL) displacements are predominantly sub-mm in magnitude in all three frequency bands and spatial components across the network, with larger residuals of up to several mm in some coastal areas. Accounting for the effects of anelastic dispersion in the upper mantle using estimates of Q from standard Earth models reduces the residuals for the M2 harmonic by an average of 0.1-0.2 mm across the network and by more than 1 mm at some individual stations. For the relatively small Mf tide, the effects of anelastic dispersion (<0.03 mm) are undetectable within current measurement error. Incorporating a local ocean-tide model for the northeastern Pacific Ocean reduces the M2 vertical residuals by an average of 0.2 mm, with improvements of up to 5 mm at some coastal stations. Estimated RMS observational uncertainties in the vertical component for the M2 and O1 tides are approximately ±0.08 mm at the two-sigma level (±0.03 mm in the horizontal components), and ±0.21 mm for the Mf harmonic (±0.07 mm in the horizontal components). For the M2 harmonic, discrepancies between predicted and observed OTL displacements exceed observational uncertainties by about one order of magnitude. None of the ocean tide and Earth model combinations is found to reduce the M2 residuals below the observational uncertainty, and no single forward model provides a best fit to the observed displacements across all tidal harmonics and spatial components. For the O1 harmonic, discrepancies between predicted and observed displacements are generally several-fold larger than the observational uncertainties. For the Mf harmonic, the discrepancies are roughly within a factor of two of the observational uncertainties. We find that discrepancies between predicted and observed OTL displacements can be significantly reduced by removing a network-uniform tidal-harmonic displacement, and that the remaining discrepancies exhibit some regional-scale spatial coherency, particularly for the M2 harmonic. We suggest that the remaining discrepancies for the M2, O1 and Mf tides cannot be fully explained by measurement error and instead convey information about deficiencies in ocean-tide models and deviations from spherically symmetric Earth structure.